Heterogeneity in High School Career and Technical Education Outcomes

Abstract

High school Career and Technical Education (CTE) has received increased attention from policymakers and researchers in recent years. This study fills a needed gap in the growing research base by examining heterogeneity within the wide range of programs falling under the broad moniker of CTE, highlighting the need for nuance in research and policy conversations that often consider CTE as monolithic. Using student-level course-taking records, unemployment insurance, and National Student Clearinghouse data, we examine outcomes including earnings, postsecondary education, and poverty avoidance. We find substantial differences for students in fields as diverse as health care, Information Technology (IT), and construction. We also highlight heterogeneity for student populations historically overrepresented in CTE, and we find large differences in outcomes for CTE students, particularly by gender.

Keywords

career development educational policy high schools policy vocational education descriptive analysis regression analyses

Introduction

Throughout the second half of the 20th century, research on vocational education in secondary schools generally found negative effects for participating students. In particular, vocational education was shown to perpetuate curricular tracking, which prevented certain student groups—particularly students with disabilities and racially minoritized students—from accessing the academically rigorous instruction that would prepare them for college and high-earning careers (Anderson, 1982; Bowles & Gintis, 1976; Oakes, 1983; Tyack, 1974). In the early 21st century, vocational education underwent a significant reinvention, highlighted in part by the shift in terminology from vocational education to Career and Technical Education (CTE), as well as increased investments by the federal government along with positioning CTE as part of a “College and Career Readiness” agenda (Dougherty, 2016; Dougherty & Harbaugh Macdonald, 2020). While many traditional vocational programs remain in place, new CTE programs tend to emphasize pathways that were explicitly designed to prepare students for postsecondary education as well careers in high-demand, high-wage areas (see Bozick & Dalton, 2013; Gottfried & Plasman, 2018).

Alongside the renewed policy interest and shifts in curricular foci of CTE at the secondary level, an emerging body of experimental and quasi-experimental research using more recent data (Bonilla, 2020; Brunner et al., 2021; Dougherty, 2018; Hemelt et al., 2019; Kemple & Willner, 2008) has enhanced our understanding of the causal effects of high school CTE programs, offering evidence of positive impacts that contrast with the earlier nonexperimental research. However, given the increasingly broad range of contexts and programs that fall within CTE, these studies are limited in that they largely treat CTE as a monolithic experience, potentially masking the extensive diversity of experiences students experience.¹

Prior research has left several opportunities to further improve our understanding of whether CTE works, for whom, and under what conditions. First, recent research has tended to focus on CTE programming within oversubscribed, specialized CTE schools, where opportunities to estimate causal impacts have arisen. However, most CTE students across the nation engage with CTE within traditional comprehensive schools or part-time centers, which may not offer the same set of experiences as whole-school models of CTE, and therefore may not produce similar effects.

Furthermore, prior research has done little to disentangle potential differences in impact among the different programs within CTE, or how students with different personal characteristics may differently experience returns to CTE. With the push to expand CTE beyond traditional vocational programs and new federal guidelines that encourage CTE to emphasize college and career readiness, any analysis of CTE today must grapple with heterogeneity across career clusters as diverse as science, technology, engineering, and mathematics (STEM); cosmetology; health care; and manufacturing. Given the push for STEM-focused CTE programs, an emphasis on how these CTE programs may lead to different outcomes than more traditional vocational programs seems especially pertinent. We also pay close attention to differences in outcomes of different student populations of interest, which is particularly relevant given the federal Perkins legislation’s explicit focus on equity in access and outcomes for different student groups.

This article begins to fill a gap in the existing research base by estimating differences in the associations between high school CTE participation and various post–high school outcomes across different career clusters and for different student populations. Using student-level administrative data from Massachusetts, we leverage factors known to be associated with selecting into CTE to observe how high school CTE program participation relates to college completion, employment, and earnings for the nine cohorts of high school students expected to graduate high school from Spring 2009 to 2017. We observe students’ high school course-taking records and outcomes between 1 and 7 years after anticipated high school graduation and find that advantages for CTE concentrators are highly heterogeneous for both college and workforce outcomes. We find that students concentrating in certain CTE fields see strong advantages in workforce, whereas students in other fields see stronger postsecondary outcomes. We also document that these advantages vary widely across the student characteristics, with students in populations that have historically been marginalized (including students with disabilities, students from lower-income families, Black and Latino students, students not attending college, and those with low tests scores) experiencing the largest benefits, along with those attending CTE-dedicated schools where the intensity of their CTE exposure may be more intensive. In particular, we find strong evidence that CTE may be a useful lever to help students avoid especially negative outcomes such as poverty and disengagement from both education and the workforce. Although not explicitly causal in nature, our estimates hold up to a series of robustness checks which suggest that even under fairly conservative assumptions these returns cannot be explained by bias alone.

In this article, we proceed as follows: We first review a brief history of the research base that motivates our work and the need to study heterogeneity within high school CTE. We then discuss the context, data, and measures we use to explore heterogeneity within CTE. We follow with a descriptive analysis, in which we focus on differences within who opts in to CTE and into different fields of study. After introducing our analytic approach, we present results in which we explore expected differences in outcomes associated with CTE by both career clusters and populations of interest. We then test limitations to our analytic approach through a number of robustness analyses. Finally, we conclude with remarks about the significance of our findings and implications for both policymakers and researchers who study CTE.

Literature Review

Given shifts in the CTE policy landscape in recent years, toward CTE as a part of a college and career readiness curriculum, an emerging body of experimental and quasi-experimental research has sought to revisit potential returns to CTE, providing some reasons for optimism for proponents. Kemple and Willner (2008) implemented a random assignment study, in which students were assigned for admission to nine oversubscribed career academies, finding that career academy participants saw no meaningful difference in postsecondary education, but did earn 11% more per year than nonparticipants over the first 8 years after high school graduation, with returns concentrated among male students (who saw a 17% increase in earnings). Hemelt et al. (2019), using more recent data from a similar admissions lottery process in one career academy in North Carolina, find an 8% increase in high school graduation rates for career academy participants. Similar to Kemple and Willner (2008), Hemelt et al. (2019) find more effects that are positive for male students, particularly when considering college enrollment. Dougherty (2018) and Brunner et al. (2021) both employ a regression discontinuity design using admission score cutoffs for CTE-dedicated high schools, with Dougherty finding a 7% to 10% increase in the likelihood of high school graduation, and Brunner et al. finding a 31% increase in quarterly earnings (again, with returns accruing primarily to male students), though evidence of null effects on college enrollment by 23 years of age. Bonilla (2020) uses a school district-level regression discontinuity on receipt of additional funding for CTE and found a reduction in high school dropout among districts that received additional funds to spend on CTE. Importantly, these impacts were stronger for girls, but schools invested principally in health services CTE programs, which are disproportionately enrolled in by females.

Overall, the emerging causal literature paints a picture of positive earnings returns, particularly for male students, with more mixed evidence of effects on postsecondary education. One limitation of all of these studies, however, is that they rely on the experiences of CTE students in oversubscribed, whole-school CTE models, which are not representative of the wide range of settings in which CTE is offered throughout different local contexts (U.S. Department of Education, 2019).

In addition to the recent experimental and quasi-experimental work, further quantitative research has enhanced our understanding of CTE in the more modern policy context and raises questions about the nuanced impact CTE may hold for participants. Kreisman and Stange (2020), for example, find evidence that participation in CTE is more widespread across academic achievement levels than in previous eras, raising doubts of whether long-standing assumptions about CTE as a “dumping ground” for low-achieving students still hold true (Kelly & Price, 2009). Kreisman and Stange also find that earning returns largely accrue to students who take upper-level CTE courses, arguing that in-depth concentration in a particular career cluster may be important for meaningful returns. Cellini (2006) finds some evidence that CTE participation increases high school graduation and 2-year college enrollment, though the 2-year college enrollment increase may be partially due to some diversion of CTE students from 4-year colleges. Other studies including Bishop and Mane (2004) and Meer (2007) also find evidence of positive returns that may vary across career cluster, though these results rely on older data from students attending school in an era before the shift from vocational education toward the modern era of CTE that aims to be more focused on academic rigor and college and career readiness.

Although an emerging base of research points to some positive benefits to CTE, there are a few general limitations that we seek to address. First, most research either uses relatively outdated data from a time when CTE plausibly operates much differently than today given recent policy initiatives (including the shift from “vocational education” to “CTE” and emphasis on applied STEM seen in Congress’ 2006 reauthorization of the Perkins Act); using more recent data, we can speak more closely to the current policy context. Second, many of the more recent studies only consider students in oversubscribed, CTE-dedicated school settings, limiting generalization to the other settings including undersubscribed CTE schools and comprehensive high schools; by incorporating statewide administrative data, we can generalize more broadly to a wide range of settings in which students engage with CTE. Finally, studies generally consider CTE as a single curricular intervention, rather than exploring differences across CTE programs (potentially in part because many studies of CTE have focused on outcomes such as attendance, high school graduation, and college-going—outcomes that may be more dependent on CTE participation overall rather than participation in a specific field of study). However, given policy interest in workforce outcomes and the rise of new STEM-focused CTE programs that were designed to be part of a college and career readiness agenda, this article seeks to explore differences in outcomes experienced by CTE concentrators across the range of career clusters within CTE.

Context

Massachusetts provides a compelling setting to study CTE participation in that it has a prominent, well-established system, a diverse range of program offerings, and a participation rate well suited for meaningful analysis. With approximately 21.5% of students across the state concentrating in CTE, there is a large sample of CTE concentrators within which we can examine several dimensions of heterogeneity. Moreover, the diversity of contexts in which CTE is offered in Massachusetts mirrors the diversity of contexts nationwide.

Slightly less than half of CTE concentrators (46%) attend CTE-dedicated schools that they are eligible to apply based on their town of residence; some of these schools are viewed as highly attractive and receive a greater number of applicants than seats available, while other CTE-dedicated schools are under-enrolled. Once in the CTE-dedicated schools, most students take exploratory courses in ninth grade, where they are exposed to multiple CTE career clusters before choosing a specific CTE pathway.

Meanwhile, just more than half (54%) of CTE concentrators take CTE courses within comprehensive schools, where some students may engage with CTE by enrolling in a specific program of study, while others may simply take CTE courses as one-off electives. Some programs are heavily funded by the state through a Chapter 74 program,² whereas others receive less funding and support. Schools have a large degree of discretion in terms of which CTE programs they offer, meaning that decisions about how to engage with CTE should be considered a combination of both student choice and school offerings.

In mixed-methods work from Massachusetts, Ansel et al. (2022) find that students’ decision (and decision-making process) to enroll in CTE was highly variable across regions, and that students and school counselors differed widely in their perceptions of CTE, though gendered patterns of participation were evident in many contexts. In some schools and regions, CTE programs were highly advertised and viewed as attractive options, while students in some regions had relatively little information about CTE in some schools and regions, with CTE in some cases viewed as options better suited for students who were struggling academically. While the decisions of students to participate in CTE (and which CTE courses to take) is an area for further study, the existence of multiple pathways also mirrors in some ways the many ways that students come to CTE nationwide.

In Massachusetts, students can concentrate in 10 career clusters by taking two or more years of courses in that cluster. While these career clusters are somewhat different than the 16 national career clusters promoted by Advance CTE (2018), there are broad enough similarities that findings in Massachusetts can help inform our thinking about heterogeneity across different career clusters nationwide. Three of the five most common clusters, construction, manufacturing, and transportation, include courses that may be thought of as more “traditional vocational” courses (in so far as they include traditional skilled trades like electrical, plumbing, construction, and auto mechanics).³ Still, a substantial portion of CTE students concentrate in clusters such as business and consumer science, communications, health care, and information technology that may break the mold of the common conception of old vocational programs, and may be more aligned with what some have called “new CTE” (Markus, 2011) and STEM-aligned pathways (Dougherty & Harbaugh Macdonald, 2020; Plasman et al., 2017).

Data

We use data from the Massachusetts state longitudinal data system (SLDS) covering cohorts of first-time ninth graders whose on-time (i.e., 4 years after entering ninth grade) graduations from high school were expected in Spring 2009 through 2017 (for most of our analyses, we focus on students from the 2009–2011 cohorts, for which we can observe 7 years of post-high school data).

The data set includes enrollment data, demographics, attendance, town of residence, Massachusetts state standardized test scores, immigrant status, disability status, and English learner status. We add college enrollment and completion data from the National Student Clearinghouse, as well as quarterly earnings data reported to the Massachusetts Department of Labor through the unemployment insurance (UI) system. We observe individual student outcomes for up to 7 years after their on-time graduation year. UI records include only taxable reported earnings for non-federal employment within the state that are eligible for unemployment benefits. While we consider those individuals with zero reported earnings within a year as non-earners in that year, this may exclude some earnings such as federal work, some seasonal work (e.g., in agriculture) that may go unreported to UI, along with self-contract employment, including some work in the “gig economy.”^4,5 The analytic sample used for most of the analyses includes 251,437 students, approximately 19.8% of whom are CTE concentrators under the state definition used for federal reporting purposes.⁶

Measures

Our primary measure of interest is whether a student completed a CTE concentration when in high school. For our purposes, this means a school identifies a student as a CTE concentrator if they are enrolled in CTE courses for two or more school years at any time during high school. This “concentrator” definition is used for federal reporting purposes, making it a meaningful designation with implications for how much Perkins funding the state receives. It also represents a substantive commitment to CTE, above and beyond any more minor exposure students would receive from taking a single CTE course as an elective credit. Moreover, many CTE clusters are explicitly designed to be completed in 2-year course sequences, with students often prepared to take licensure/certification exams, or to receive industry or state-recognized credentials after 2 years of CTE courses. In the analyses in which we consider the advantages for CTE concentrators in specific career clusters, we count only those students taking two or more years of courses in that cluster to be cluster concentrators (e.g., health care concentrators, construction concentrators). For those students who completed two or more years of CTE, but not within a single cluster (sometimes referred to as CTE “dabblers”), we include them as CTE concentrators, but not as concentrators in any one cluster for the cluster-specific analyses. These “dabblers” receive substantial exposure to CTE courses, but are mostly enrolled in comprehensive high schools where some CTE programs can be less defined (as opposed to CTE-dedicated schools), and some simply become dabblers by taking multiple CTE courses as electives without necessarily making conscious decisions about their CTE concentrator status.

Our key outcomes of interest are college enrollment, college completion, earnings, employment, and economic outcomes that are associated with economic dependence on the state (poverty, and being neither enrolled in college nor employed). We define these outcomes as follows. First, we define enrolling in any college as a binary indicator equal to 1 if individuals are ever observed enrolling in a 2- or 4-year college after completing high school. We also create separate indicators to capture whether students graduate from a 2-year college, a 4-year college, or complete a certificate or degree at either type of institution. For labor market outcomes, we examine total annual earnings in each of the first 7 years after expected completion of high school, as well as binary indicators of whether individuals earned at or above the inflation-adjusted federal poverty level at each of these time periods. Our final outcomes of interest are whether students are neither employed nor enrolled in college (NEET) during the first 7 years after expected completion of high school, and whether an individual earned enough money to clear the federally defined threshold for poverty for a household size of one. These latter sets of outcomes help us understand whether students are able to avoid outcomes known to be associated with larger negative personal and social costs.

Descriptive Analyses

Heterogeneity within CTE occurs on two clear dimensions that we explore here—the characteristics of students who become CTE concentrators relative to non-CTE students, and the characteristics of students across career clusters.⁷ Table 1 and Figure 1 highlight the starkness of these differences. Echoing work from other settings (Dougherty, 2018; Dougherty et al., 2018; Plasman et al., 2020, among others), Table 1 shows that CTE concentrators are less likely to be females, more likely to be lower-income, and more likely to be English language learners than non-concentrators. In terms of racial and ethnic identity, Latino students are especially overrepresented and Asian students underrepresented among CTE concentrators. CTE concentrators score well below the state average on eighth grade standardized tests and are nearly 13 percentage points less likely to attend and graduate from college (especially 4-year colleges) than their non-CTE peers.⁸

Table 1

Descriptive Statistics for CTE Concentrators and Non-Concentrators

Student characteristic	CTE concentrators	Non-concentrators
Female	0.43	0.50
Female	(0.50)	(0.50)
Free/reduced lunch	0.56	0.37
Free/reduced lunch	(0.50)	(0.48)
Students with disabilities	0.24	0.18
Students with disabilities	(0.43)	(0.38)
Immigrant	0.04	0.05
Immigrant	(0.20)	(0.22)
English language learners	0.07	0.07
English language learners	(0.26)	(0.26)
Latino	0.20	0.15
Latino	(0.40)	(0.35)
Asian	0.04	0.06
Asian	(0.19)	(0.23)
Black	0.11	0.10
Black	(0.31)	(0.30)
White	0.70	0.72
White	(0.46)	(0.45)
Grade 8 math score (Std.)	−0.32	0.08
Grade 8 math score (Std.)	(0.79)	(0.93)
Grade 8 ELA score (Std.)	−0.58	−0.31
Grade 8 ELA score (Std.)	(0.80)	(0.93)
Grade 8 attendance rate	0.96	0.95
Grade 8 attendance rate	(0.05)	(0.06)
Attend regional vocational school	0.44	0.00
Attend regional vocational school	(0.50)	(0.06)
On-time high school graduation rate	0.81	0.71
On-time high school graduation rate	(0.39)	(0.45)
Attend 2-year college	0.40	0.29
Attend 2-year college	(0.49)	(0.46)
Attend 4-year college	0.36	0.54
Attend 4-year college	(0.48)	(0.50)
College graduate	0.28	0.44
College graduate	(0.45)	(0.50)
Took some CTE, but not a CTE concentrator	—	0.10
Took some CTE, but not a CTE concentrator	—	(0.30)
Observations	49,691	201,746

Note. Analytic sample includes first-time ninth graders in cohorts that would have graduated on time from public high schools in the Spring years of 2009 through 2011. Students are considered to be a “CTE concentrator” if they are enrolled in CTE for at least two academic years. CTE = Career and Technical Education; ELA = English Language Arts.

Figure 1.

CTE concentrators’ characteristics in each career cluster compared with statewide student population.

We present in Figure 1 the over- or underrepresentation of select student characteristics, relative to the statewide average (represented by the horizontal red line in each panel) by cluster and show clear variation. Perhaps the most striking differences relate to gender. Construction concentrators are 36 percentage points more male than the statewide average, with male students also widely overrepresented in the transportation, manufacturing and technology, and Information Technology (IT) clusters. In contrast, male students are 43 percentage points less likely to concentrate in education than the state average, and also highly underrepresented in health care and business and consumer sciences.

Figure 1 (see also Supplementary Table A2 in the online version of the journal) also highlights that prior academic performance differs across clusters. Students scoring in the lowest quintile of eighth grade test scores are overrepresented in every cluster, though low-scoring students are particularly present in transportation, hospitality and tourism, and construction. Substantial differences in selection into CTE also exist across clusters for lower-income students, students with disabilities, and Black and Latino students. In addition, clear in Figure 1 is that while CTE concentrators as a whole are less likely to attend college than the statewide average, this varies by career cluster. In some clusters (health care, education), students are descriptively somewhat more likely to lead to college than the statewide average, whereas in others (construction, transportation) students are far less likely to enroll in college than the statewide average.

These underlying differences in the characteristics of students who become concentrators in the different career clusters present a compelling case that we might consider each cluster as a distinct intervention, rather than one single program, broadly labeled as CTE. Because students who have access to and/or choose to opt into CTE vary so widely across clusters, it appears that students themselves may view the clusters quite differently. Thus, the construction of potential counterfactuals should account for those differences in models and estimate different impacts by cluster.

Figure 2 displays descriptive trends and differences in earnings across the career clusters, separated by whether students concentrated in CTE and also whether they attended college. As expected, students who attend college earn more than students not attending college, especially as more time passes after high school. In certain fields tightly connected with college-going (e.g., health care; see Figure 1), CTE students who attend college earn more than non-CTE college students, suggesting that in pathways tied explicitly to postsecondary education (e.g., nursing) CTE may be especially advantageous. Among students not attending college, CTE concentrators in every cluster attain higher earnings on average throughout the first 7 years after their expected high school graduation, with this descriptive advantage for concentrators in some clusters (particularly in the traditional trades such as engineering and manufacturing, construction, and transportation) increasing over time. On the whole, college-goers’ earnings increase rapidly in Years 5 through 7, as many college-going students enter the workforce. However, in these same trade fields, CTE concentrators not attending college continue seeing competitive earnings to those attending college, even 7 years after high school. In Supplementary Figure A2 in the online version of the journal, we also present descriptive differences in the rates of college-going and degree attainment, compared with the statewide averages (represented by the long red-dashed lines) demonstrating substantial differences across cluster in how likely students are to attend college; across every cluster, however, CTE concentrators are less likely to complete a college degree than the statewide average, though, again, this varies widely.

Figure 2.

Annual earnings for CTE concentrators in each cluster compared with non-CTE concentrators, through 7 years after high school.

Analytic Approach

Because student self-selection is endemic to high school curricular choice, descriptive analyses—while informative—almost certainly obscure the role of CTE in helping students achieve certain outcomes. While a regression-based approach is prone to bias from unobserved variables that may predict selection into CTE, our approach allows us to take advantage of a statewide database in which students engage with CTE in vastly different contexts. This approach also allows for stronger generalizability, as we are able to consider CTE in both CTE-dedicated settings and comprehensive school settings, and across a wide range of career clusters, mirroring the many different ways CTE is offered across American public schools.

While the primary aim of this analysis is to explore heterogeneity within CTE, we first establish the credibility of our analytic approach by fitting a model to compare student outcomes for students who are observably similar and had access to a similar set of school and curricular options. We specify our main model as follows:

Y_{i c t} = β_{1} C T E_{i} + X_{i}^{'} γ + π_{c} + τ_{t} + ϵ_{i c t} .

(1)

Here, Y_ict is a generic outcome for student i, in cohort c and town t. The key predictor CTE is equal to 1 if student i is a CTE concentrator (0 otherwise), $X_{i}^{'}$ is a vector of student-level covariates including demographic characteristics and eighth grade test scores and attendance, π_c represents fixed effects for entering cohort, and τ_t represents fixed effects for town of residence. Errors are heteroscedasticity robust and clustered by the town of residence. In all models, β₁ is the coefficient of interest and represents the average population difference in a given outcome associated with CTE concentration, relative to otherwise similar non-concentrators. We also consider an alternate counterfactual group, students who take a single year of CTE but do not concentrate, which we discuss in the section “Limitations and Tests for Robustness.”

While we cannot rule out the presence of unobserved factors predicting selection in CTE, and accordingly use non-causal language when interpreting our estimates, our models include a rich set of controls for student-level demographic information, eighth grade school attendance rates, and eighth grade performance on state assessments (both mathematics and English Language Arts), which account for unobservable characteristics that would influence both eighth grade academic performance and selection into high school CTE. Cohort and town of residence fixed effects account for differential labor market trends and access to CTE offerings.

To demonstrate the merits of our approach, we fit initial models that show the stability of our estimates once we include an increasing number of controls. Following the argument in Altonji et al. (2005), the stability of these estimates across the more saturated models provides evidence that we have accounted for the most egregious sources of potential bias. We also apply Oster’s (2019) approach by estimating how large remaining unobserved selection bias would have to be to nullify our estimates (discussed in section “Limitations and Tests for Robustness”), and find that unobserved bias would have to substantially outweigh observed explained variation for true impacts to be zero.

Before examining the different dimensions of heterogeneity for our different outcomes, in Table 2, we present our estimates of the relationship between CTE concentration and one outcome (cumulative earnings over the first seven post-high school years) using five specifications of model for students concentrating in each of the 10 career clusters (1).⁹ Each specification sequentially adds controls to better isolate any difference in outcomes that might be associated with CTE concentration. By adding year and town of residence fixed effects, in Model 2, we highlight that a portion of these differences can be explained by the contexts in which students live, the years in which they enter high school, and the schools they can attend. This also highlights that access to CTE offerings (as dictated by what school someone attends) plays an important role in driving unconditional differences in outcomes but does not fully account for differences.

Table 2

Regression-Adjusted Estimates for Each CTE Career Cluster Concentration on Select Cumulative Earnings Over the First 7 Years Post-High School

Career cluster	I	II	III	IV	V
Health care	31,220.65	35,973.3	31,693.38	35,324.68	32,570.59
Standard error	2,160.071	2,483.01	2,176.249	2,475.084	2,079.79
Observations	205,316	205,316	205,316	205,316	205,316
Education	14,604.71	17,571.52	14,532.6	20,527.64	17,996.69
Standard error	1,666.653	1,744.966	1,654.109	2,219.148	1,854.856
Observations	204,051	204,051	204,051	204,051	204,051
Information technology	16,570.46	16,889.72	12,835.86	18,510.43	13,098.68
Standard error	2,301.413	2,155.278	2,056.111	2,331.869	2,150.374
Observations	204,059	204,059	204,059	204,059	204,059
Arts and communication	7,233.973	9,519.232	6,024.094	9,635.332	6,126.694
Standard error	1,448.891	1,369.872	1,458.307	1,633.192	1,418.954
Observations	206,117	206,117	206,117	206,117	206,117
Business	21,899.78	24,056.06	20,526.65	23,014.96	20,407.06
Standard error	1,942.044	1,647.925	1,488.458	1,761.041	1,365.647
Observations	207,911	207,911	207,911	207,911	207,911
Agriculture	14,735.02	16,066.52	13,574.41	15,587.56	13,869.62
Standard error	2,681.871	2,471.453	1,972.986	2,842.125	2,330.243
Observations	203,734	203,734	203,734	203,734	203,734
Hospitality	12,725.73	16,521.58	13,167.3	15,818.11	14,753.18
Standard error	1,652.558	1,602.283	1,555.519	1,682.908	1,530.081
Observations	205,503	205,503	205,503	205,503	205,503
Engineering and manufacturing	29,075.82	28,611.39	25,195.43	31,577.66	26,108.08
Standard error	1,963.436	1,936.152	1,744.748	1,907.752	1,502.024
Observations	207,377	207,377	207,377	207,377	207,377
Construction	49,386.04	48,550	45,119.36	49,721.55	44,761.62
Standard error	1,788.445	1,791.622	1,668.635	1,847.146	1,611.037
Observations	209,749	209,749	209,749	209,749	209,749
Transportation	37,615.57	40,216.34	36,539.9	39,603.03	37,016.74
Standard error	1,759.247	1,657.522	1,580.524	1,884.191	1,632.641
Observations	205,982	205,982	205,982	205,982	205,982
Controls for demographic characteristics	No	No	Yes	Yes	Yes
Controls for eighth grade assessments and attendance	No	No	No	Yes	Yes
Fixed effects for cohort and town of residence	No	Yes	No	No	Yes

Note. Estimates are the coefficient associated with CTE concentration in each cluster, specified by row, on students’ cumulative earnings over the first 7 years after expected high school graduation. Model I includes only an indicator of CTE concentration and the outcome of interest. Model II adds cohort and town of residence fixed effects, with errors clustered by town of residence. Model III includes controls for gender, race and ethnicity, lower-income status, English language learner status, immigrant status, and disability status. Model IV adds eighth grade school attendance rates, and eighth grade performance on state assessments (both mathematics and English Language Arts) to demographic controls. Model V includes both fixed effects and all controls. Analytic samples include first-time ninth graders in cohorts that would have graduated on time from public high schools in the Spring years of 2009–2011. Students are considered to be a “CTE concentrator” if they are enrolled in CTE for at least two academic years. Comparison students are those who were never enrolled as a CTE student. CTE = Career and Technical Education.

Adding student demographic characteristics (Model 3) reduces the magnitude of the CTE concentration estimate across all clusters, suggesting that observable student characteristics do not differentially explain differences in outcomes across clusters. Model 4 adds controls for eighth grade assessments and eighth grade attendance rates, which allow us to consider students within the context of demonstrated academic performance prior to any engagement with CTE in high school. As with demographics, adding test scores as controls changes the relationship between CTE concentration and total earnings in similar ways across clusters. Finally, Model 5 includes all controls from earlier models, as well as town and cohort fixed effects. Across all career clusters, the direction and significance of the estimates remain consistent across model specifications, lending confidence to the inference that there is some persistent contribution of CTE concentration to earnings. While there are some differences in how the various specifications change the estimates across the different clusters, these differences are relatively minor, with estimates in Model V generally no more than 10% larger or smaller than the unconditional estimates in Model I. The largest increase in returns from Model I to Model V are in education and hospitality, suggesting that there may be more negative selection into these clusters. While these clusters may have some negative selection, they also see some of the weakest returns, alleviating concerns that we are overstating their earnings advantage. Conversely, after conditioning on these factors, returns are somewhat more modest for the arts and IT, suggesting that there may be positive selection into these clusters (and that we may be somewhat overstating our findings). While there are some differences in how selection driven by observables may not be identical across clusters, the relatively minor sensitivity to changes in the model are small enough that they do not change the overall takeaways of our analysis. Thus, throughout the rest of the article, we present results using the specification from Column 5 (our fully specified model in Equation 1).

Results

Postsecondary Outcomes

In Figure 3, we present estimates of the relationship between CTE concentration and postsecondary outcomes (also presented in Supplementary Tables A4 and A5 in the online version of the journal). Each panel of Figure 3 presents β₁ for the overall population of students, male and female students, and for several populations of interests; in particular, we focus on student populations for whom there have historically been concerns about inequitable tracking into CTE. For reasons of sample size and statistical power, we present results for Black and Latino students together, though findings are similar for both populations. Moreover, we focus on populations that have been historically underrepresented in higher education and have faced lower earnings outcomes, and thus are of particular interest to policymakers, and researchers focus on CTE. Throughout the article, overall results are presented first, with results then presented from left to right in order of most to least likely (based on the descriptive evidence above) to attend college. Vertical bars on each coefficient result represent 95% confidence intervals. Intervals not crossing zero indicate statistical significance at alpha = .05 level. Throughout the article, differences and advantages or disadvantages for CTE students that we discuss can be interpreted as statistically significant at the 95% level or better, unless otherwise noted. Looking first at the top-left panel of Figure 3, CTE concentration predicts essentially no difference on the extensive margin of attending any college; however, this estimate varies by population. Female students see a moderate (3.8 percentage point) increase in their overall rate of college-going, while male students see a minor, but statistically insignificant decrease. Conversely, Black and Latino students see a large increase in their likelihood of attending college (8.9 percentage points), as do students eligible for free or reduced-price lunch. CTE concentration is also associated with a decrease in overall degree attainment (see Supplementary Figure A3 in the online version of the journal), though, notably, this relationship is not significant in population groups that are currently underrepresented in college-going, particularly Black and Latino, students with low test scores, and those eligible for free and reduced-price lunch.¹⁰

Figure 3.

CTE concentrators’ college attendance outcomes compared with similar non-concentrators, by populations of interest and career cluster.

Figure 3 also presents results specifically for 2- and 4-year college enrollment. Here, CTE is associated with an increase in attending a 2-year college and smaller or insignificant decreases in 2-year college-going degree attainment across all subpopulations. However, CTE is associated with lower rates of attendance at 4-year colleges (although, again, the negative associations are not significant for Black and Latino students and students with disabilities). Overall, Figure 3 highlights a picture in which CTE is associated with a modest overall decrease in college-going and, in particular, attainment. While our approach cannot definitively speak to whether CTE leads some students to substitute away from 4-year colleges and into 2-year colleges, this pattern of substitution at the intensive margins of college enrollment would be consistent with our estimates.

The right hand panels of Figure 3 (also in Supplementary Table A5 in the online version of the journal) explore the same education outcomes as above, but rather than comparing outcomes for CTE concentration more generally across different student populations, we now present differences of outcomes accruing to CTE concentrators in each specific career cluster. For example, to estimate anticipated advantages from concentration in the health care cluster, we compare health care concentrators with non-CTE students who were otherwise similar on observable characteristics. For cluster-specific analyses here (and throughout the article), concentrators in clusters other than the one under study are excluded, which allows us to examine the expected difference for students who become CTE students in a particular career cluster, compared with students who do not concentrate in CTE. Again, we arrange results from the clusters where students are descriptively most likely to attend college (health care, education) to least likely (transportation, construction). Interestingly, even after accounting for student and local characteristics, the clusters with the highest college-going rates also see the strongest increases in the probability of college attendance. The differences between the advantages for health care and education concentrators (15.1 and 13.0 percentage points) and the disadvantages for transportation and construction (−14.9 and −12.7) are striking. For transportation and construction, this is driven almost entirely by large decreases in the likelihood of attending 4-year colleges (−17.7 and −16.6). In addition, there are several clusters in the center where students experience little to no change in their likelihood of attending college. In terms of degree attainment, decreases in college-going are especially notable for the less college-going clusters (see Supplementary Table A5 in the online version of the journal). Most of the clusters are associated with an increase in 2-year college-going, and in some cases, modest increases in 2-year college completion. Finally, some clusters (most notably health care, IT, and education) see large increases in overall and 2-year college attendance without an equivalent decrease in 4-year college attendance, suggesting that these clusters (which often require additional education to be completed at least at the 2-year college level) may be inducing some students to attend 2-year colleges who otherwise might not have pursued postsecondary education. Given that some career clusters (e.g., health care) have clearly aligned paths to the community college level (i.e., nursing programs), the strong relationship with 2-year college attendance is notable and likely speaks to the design of the pathways.

Earnings

While policymakers have increasingly pointed to postsecondary education as an important intended outcome of CTE, another long-standing goal for students is to position themselves for higher earnings. We turn now to the question of how students may expect to benefit financially from their engagement with CTE.

Figure 4 (see Supplementary Table A6 in the online version of the journal) displays the predicted impact of CTE concentration for the first 7 years after high school graduation.¹¹ Overall, CTE concentration is associated with a large increase in initial earnings (US$1,792 in the first year after high school) that persists even 7 years after high school (US$3,359 in annual earnings). Figure 4 also presents clear differences in who sees positive earnings advantages from CTE. Advantages are especially strong and persistent for students who never attend college within the first 7 years after high school (whom we refer to as “No College”), with CTE No College students earning US$6,053 more in the seventh post-high school year than otherwise similar “No College” peers who are not CTE concentrators. Echoing results from prior studies (Brunner et al., 2021), male students see larger differences attributed to CTE concentration, whereas female students see more modest advantages that quickly diminish over time. Moreover, CTE is associated with an increase in earnings for several of the student populations who have been historically marginalized, especially students with disabilities, as well as lower-income students, Black and Latino students, and students with the lowest prior achievement scores.

Figure 4.

CTE concentrators’ annual earnings advantage compared with similar non-concentrators, by populations of interest.

Figure 5 (see Supplementary Table A8 in the online version of the journal) presents the relationship between earnings and CTE concentration, here disaggregated by career cluster.¹² The heterogeneity in these results across cluster are even greater than the differences across student populations presented in Figure 4, lending support to the hypothesis that cluster selection is crucial in determining whether and how they might expect to benefit from CTE. Looking across cluster, the strongest predicted increase in earnings is associated with the construction, transportation, manufacturing and technology, and health care clusters, while students in hospitality, agriculture, and communications see little to no predicted benefit in their earnings, especially as students are further removed from high school graduation. In most clusters, the positive association with earnings begins to subside in Years 5 through 7 (likely as college-goers who may serve as the counterfactual re-enter the workforce); still, it is noteworthy that in the career clusters with the highest predicted advantages (especially health care and construction), the earnings advantages remain large (though in transportation, the advantage noticeably declines by 7 years after high school).

Figure 5.

CTE concentrators’ annual earnings advantage compared with similar non-concentrators, by career cluster.

Poverty and Disengagement

While CTE may be thought of as a way to increase earnings and education, it has also often been thought of as a tool to reduce the most adverse outcomes. This is of particular importance for students who face social and economic disadvantages and inequitable access to services that may make them vulnerable to negative outcomes after high school. CTE may therefore also be evaluated by the extent to which it reduces students’ likelihood of living in poverty, or of being Neither Employed nor in Education or Training (NEET, or disengaged).

We first turn to a measure of disengagement that combines both education and earnings to assess the extent to which a young adult is NEET. We consider someone to be NEET if they fail to either earn above the single-person poverty threshold¹³ or to be enrolled in any postsecondary institution in that year. In Figure 6 (see Supplementary Table A10 in the online version of the journal), we present some evidence to suggest CTE may be a tool to reduce overall disengagement, particularly among students who do not go to college. Figure 6 highlights that student CTE concentrators are 7.8 percentage points less likely to be disengaged from both education and the workforce 7 years after high school. Among students not attending college, we find evidence that CTE may be especially useful in mitigating the risk of being NEET in the earliest years after high school, though a sizable advantage (14.1 percentage points) remains even 7 years after high school. Advantages are also notable among many of the specific populations examined. In Supplementary Figure A8 in the online version of the journal, we illustrate differences by career cluster, finding that while differences in the magnitude of the advantage, CTE concentrators in every cluster are less likely to be NEET than otherwise similar non-CTE students.

Figure 6.

CTE concentrators’ difference in likelihood of being neither employed nor in education or training (NEET) compared with similar non-concentrators, by populations of interest.

In Supplementary Figures A9 and A10 in the online version of the journal, we also present evidence of CTE concentration’s relationship with poverty avoidance, with CTE students 8.4 percentage points more likely to avoid poverty 7 years after high school than we might expect (and a nearly 14.1 percentage point advantage among students not attending college). Again, male CTE concentrators see stronger advantages than female concentrators, and CTE concentrators across all the populations examined here see a lower likelihood of poverty, lending strength to the argument that CTE may help students avoid poverty, at least in the early years of their adulthood. Moreover, as seen in Supplementary Figure A8 in the online version of the journal, the positive relationship between CTE and poverty avoidance holds across all career clusters, with CTE concentrators substantially more likely to at least earn above the poverty threshold than other observable factors would suggest, even 7 years after high school when most college-going students would have re-entered the workforce.

School Setting

An important feature of CTE in Massachusetts is that it is offered in both comprehensive high school settings and in CTE-dedicated schools of choice. These CTE-dedicated schools differ substantially in their perceived quality and student/family demand, but in CTE-dedicated schools, all students concentrate in CTE, and all students opted-in to attend that school. These schools might then represent a more intensive exposure to CTE than concentrators are likely to receive at comprehensive schools, where many students become CTE concentrators through focused elective course-taking. While these two settings mirror two common ways CTE is offered nationwide, we might worry that results are driven primarily by a particular type of school—especially a CTE-dedicated school that has been the focus of most recent quasi-experimental CTE research. In Figure 7, we disaggregate results for students attending CTE-dedicated school and those attending comprehensive high schools. Here, it is clear that CTE concentrators in CTE-dedicated schools, at levels differing by career cluster, see a substantially larger advantage compared with non-concentrators. However, by definition, these concentrators are compared with otherwise similar students at comprehensive high schools (see the section “Limitations and Tests for Robustness,” for a discussion of why these school type-specific analyses introduce the potential for additional selection bias and how we address this).¹⁴ Yet, while the earnings advantages are greater for students at CTE-dedicated schools (echoing results seen in Connecticut from Brunner et al., 2021), it remains relevant that earnings advantages persist for CTE students at comprehensive schools, especially in the career clusters with the highest returns, albeit at more modest magnitudes.

Figure 7.

CTE concentrators’ annual earnings advantage compared with similar non-concentrators, by career cluster and school setting.

Limitations and Tests for Robustness

A key limitation of these findings involves the possibility of omitted variable bias, particularly selection bias associated with student sorting into CTE (or into specific career clusters). While the inclusion of pre-high school assessment scores and fixed effects work to alleviate these concerns, we follow the example of Oster (2019) by examining the extent of selection on unobservable characteristics that would be needed to invalidate our results. We present the results of this test in Supplementary Table A14 in the online version of the journal, using both R_max proposed by Oster of R_max = 1.3R, and a more conservative R_max = 2R. The coefficient bound on each outcome of interest tells us the range of possible coefficients on β₁ (CTE Concentration) from a model with no unobserved bias to potential models with unobserved characteristics explaining 30% as much selection as our observed characteristics. If zero does not fall within this range, it tells us that unobserved bias would need to explain more than 30% as much as observed characteristics. The bias parameter δ represents how many times larger unobserved factors would need to be than observed characteristics to nullify the results. We next take a similar approach but with R_max = 2R. Given that no coefficient bounds include 0 and all bias δs are greater than 1,¹⁵ we can conclude that selection on unobservables would need to be larger than on observables to invalidate results (and in many cases, far larger).

Concerns that students who never take CTE courses are not an appropriate counterfactual group for these analyses may also arise; in particular, if our observed characteristics do not account for differential selection into the various career clusters. In Supplementary Tables A15 and A16 in the online version of the journal, we compare students with a new counterfactual group, students who took 1 year (but no more than 1 year) in the same cluster, relying on the assumption that students taking 1 year of agriculture classes, for example, showed some interest in agriculture and might be a more suitable comparison. These results show a mix of similar and different findings; however, we posit that this is actually not an appropriate counterfactual. Students taking only a single year in a career cluster are relatively rare and exceedingly rare at the CTE-dedicated schools, in which CTE concentration is required. As such, this counterfactual primarily consists of students at comprehensive high schools mainly taking a CTE course as an elective, rather than indicating a more substantial interest in CTE.

As an additional robustness check against concerns of selection bias driving our results, we also use propensity score matching, which allows us to compare outcomes for students who had a similar likelihood of concentration in CTE (as measured by observable characteristics). One advantage of this approach is that it limits our sample only to those students within the range of common support (Abadie & Imbens, 2006; Rosenbaum & Rubin, 1983). In this case, we exclude students who were both extremely unlikely to become CTE concentrators, as well as those who were nearly certain to become CTE concentrators. We present estimates of the relationship between CTE concentration and earnings 7 years after high school using propensity score matching in Supplementary Table A23 in the online version of the journal, finding similar results to our primary model, which strengthens our findings.

While Figure 7 demonstrates differences in earnings advantages by school setting, there may be unobserved selection bias in these results due to the opt-in nature and selective application processes at many CTE-dedicated schools. To reduce concerns of bias driven by who attends CTE-dedicated schools, we must consider the returns to CTE outside of any decisions about high school attendance. To do this, we estimate models among only those students residing in towns that were not eligible for a CTE-dedicated school. Although estimates from this specification are not identical to the full sample, the sign and significance of these results mirror the full sample, emphasizing that our main results are not solely driven by CTE-dedicated schools. In fact, in Supplementary Table A17 in the online version of the journal, we present results of only those students residing in towns eligible for a vocational/technical school and find very similar results, whereas in Supplementary Table A18 in the online version of the journal, we show outcomes for students in towns not eligible for CTE-dedicated schools. Echoing the findings from Figure 7, the direction and significance of the outcomes are similar, with somewhat smaller magnitudes for students not eligible to attend CTE-dedicated schools. In Supplementary Tables A19 and A20 in the online version of the journal, we approach this idea in a different way, looking only at those students who attend comprehensive schools (i.e., did not attend a CTE-dedicated school). While this removes one key mechanism through which CTE may matter in Massachusetts (selection of high school), Supplementary Tables A19 and A20 in the online version of the journal make clear that the associations between CTE and later outcomes largely hold (albeit diminished) even at comprehensive schools.

Finally, our analyses are subject to limitations of UI data. First, not all earnings are reported through UI data, including “gig economy” work and self-contract employment (Collins et al., 2019). Another limitation of our UI data is that we are unable to observe earnings that occur outside the state of Massachusetts, a concern potentially exacerbated by the relatively small geographic size of the state and borders with six others. This may be especially important for higher-income earners and college-goers (Foote & Stange, 2019). Given that some of those individuals reporting US$0 earnings are likely earning, just not in a way captured by our UI data, we also fit models in which we exclude all those not reporting earnings in a given year. After removing these zeros from analyses, the point estimates are moderately lower across the clusters, suggesting that while some combination of actual unemployment and employment not eligible/reported for unemployment may partially drive the advantage we see for CTE concentrators, the comparison with non-reported earnings does not fully account for the CTE advantage. To examine the extent to which our estimates are biased by students crossing into other states for work, we also re-ran our earnings-related models without students from towns bordering another state, who due to proximity may be most likely to work across a state border (and be missing from UI records). After excluding these students, our results across all career clusters and student populations are nearly identical (see Supplementary Tables A21 and A22 in the online version of the journal), suggesting that students working out of state are not a major source of bias. Moreover, we see especially large advantages for CTE students among two populations less likely to move out of state for work, lower-income students and students with disabilities, which provides additional confidence that our results would hold even after accounting for unobserved out-of-state work.

Conclusion and Discussion

One challenge for evaluating the success of CTE programs is that it can be difficult to identify optimal outcomes. Some may view academic and college preparation as a primary goal, particularly given the economy’s increasing reliance on jobs that require postsecondary education (Carnevale et al., 2015; Holzer & Baum, 2017). Others may argue that CTE should prepare students for high-wage, high-growth jobs that they are qualified for immediately after their high school CTE experience. Ideally, CTE programs might prepare students for both college and career, as both federal and Massachusetts policy has worked to emphasize in recent years. One key finding from these analyses is that different CTE programs appear to help students attain different positive outcomes to varying degrees.

Figure 8 demonstrates that the relationship between CTE and different student outcomes vary across student populations. Black and Latino, lower-income concentrators, and those scoring poorly in eighth grade tests all see positive anticipated advantages in both dimensions—income and postsecondary enrollment. Overall, we find evidence that CTE is associated with a higher students’ predicted earnings even 7 years after high school, along with almost no change in postsecondary enrollment (at least in the aggregate). For male students, the change in predicted outcomes are especially stark, with both the largest predicted increase in earnings and the largest decrease in the likelihood of college attendance, likely in part because of the different career clusters they select. With many studies now finding evidence of stronger economic returns to CTE for men than women across a number of contexts (Brunner et al., 2021; Hemelt et al., 2019; Kemple & Willner, 2008; see also Dougherty & Ecton, 2021, for a discussion of this trend across an international context), these results could offer one explanation—with women disproportionately sorting into CTE fields (e.g., health care, education) that lead students into postsecondary education, as opposed to men, who disproportionately opt into fields (e.g., construction, transportation) that are far more likely to set students up for direct entry into the workforce. These findings suggest that selection of field of study within CTE are a key driver of the differences in outcomes by gender, though more work should be done to unpack these gender gaps—particularly the extent to which gender gaps persist even after entry into the workforce.

Figure 8.

Career and technical education concentrator differences in education and earnings outcomes, by populations of interest.

As Figure 9 highlights, some career clusters are more positively associated with higher earnings, whereas others are more associated with higher rates of postsecondary success. Some clusters, such as health care, education, and IT, perform especially well on both dimensions. Other clusters, such as construction and transportation, might represent a trade-off for students, in which students can expect higher earnings, but a lower likelihood of college attendance. These differences may be by design. Some programs such as health care and education neatly tie into a postsecondary pathway and may receive explicit preparation and encouragement to continue in those programs. Other career clusters, such as construction and transportation, may be more explicit about encouraging direct entry into the workforce through apprenticeship and school-to-work programs. Encouragingly, no clusters fall in or even near the bottom left quadrant of Figure 9; all clusters point to positive outcomes on at least one dimension.

Figure 9.

Career and technical education concentrator differences in education and earnings outcomes, by career cluster.

By considering a wide range of outcomes, different relationships across the career clusters, and different anticipated advantages and disadvantages to CTE across student populations, we present a nuanced picture of the wide range of heterogeneity within CTE. For advocates of CTE, these results offer evidence that CTE is associated with positive labor market outcomes, particularly for male students and students from historically marginalized backgrounds. Some of the labor market advantages may be partially driven by some students foregoing college, particularly at the 4-year college level, at least in the first years after high school. However, earnings advantages persist for CTE concentrators even 7 years after high school, at which point most college attenders will have re-entered the workforce. Given the nature of many CTE programs, it might make intuitive sense that some CTE concentrators may be more likely to develop the skills and professional network that allows them to enter the workforce immediately after high school. For some, postsecondary education may come later, as they are better able to afford college and as they need additional education and training to advance in their careers.

While advantages persist through 7 years for CTE concentrators as a whole, it is important to note that this also varies by cluster, with clear evidence of diminishing returns over time for students in clusters including arts and communication, hospitality, agriculture, and transportation. Strong early career earnings can help young adults build a foundation of financial stability; however, many CTE students will eventually need to return to postsecondary education or training to maintain strong earnings levels (Holzer & Baum, 2017). Following similar evidence of diminishing returns to CTE from Europe (Brunello & Rocco, 2017; Hanushek et al., 2017), policymakers should pay special attention to the extent to which CTE students are able to adapt to new work over time by ensuring CTE coursework balances the skills needed for specific jobs and a focus on developing students’ general skills (critical thinking, problem-solving, reading, etc.) that are especially important in a rapidly changing economy (see Autor, 2019; Lazear, 2009). If CTE coursework does not incorporate general skill development, these courses may be ripe for re-examination.

Finally, we find suggestive evidence that CTE may be especially beneficial as a stopgap to prevent some of the worst possible outcomes for students—poverty and disengagement, as CTE is associated with a decreased likelihood of earning below the poverty line and a decreased likelihood of being completely disengaged from both education and employment. Given that individuals earning below the federal poverty threshold and not engaged in education are far more likely to rely on government assistance programs, this outcome may be of particular policy relevance given the financial implications. These advantages to CTE are especially strong for students who do not enter college in the first 7 years after high school. As policymakers consider ways to help their most vulnerable students avoid these negative post-high school outcomes, CTE may be an especially attractive option. Moreover, while some of the traditional vocational career clusters such as construction, transportation, and manufacturing and technology are associated with negative college outcomes, these clusters are also associated with the strongest pay-offs in terms of expected earnings.

As states and districts consider their menu of CTE offerings, these findings have important implications for researchers and policymakers. Importantly, CTE outcomes are different for different types of CTE career clusters and across different student populations. Some career clusters may offer stronger benefits than others, while some students might be more poised to realize those benefits than others. In many cases, CTE may represent a set of trade-offs between early career earnings and postsecondary education, though these trade-offs manifest themselves in heterogeneous ways. Ultimately, our findings encourage a re-framing of conversations around CTE that moves beyond the standard consideration of CTE as a single, monolithic curricular policy, to one that embraces the substantial heterogeneity across the many different student populations and programs under the broader CTE umbrella.

Supplemental Material

sj-docx-1-epa-10.3102_01623737221103842 – Supplemental material for Heterogeneity in High School Career and Technical Education Outcomes

Supplemental material, sj-docx-1-epa-10.3102_01623737221103842 for Heterogeneity in High School Career and Technical Education Outcomes by Walter G. Ecton and Shaun M. Dougherty in Educational Evaluation and Policy Analysis

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Walter G. Ecton

Notes

Authors

WALTER G. ECTON is an assistant professor of education policy at Florida State University. His research focuses on the intersections between high school, higher education, and the workforce, with a specific focus on Career and Technical Education (CTE) in both the K–12 and higher education spaces. He is particularly interested in pathways through education and into early career for students who are historically and currently marginalized in traditional academic settings.

SHAUN M. DOUGHERTY is an associate professor of public policy & education at Vanderbilt University’s Peabody College of Education & Human Development. His work focuses on education policy analysis, causal program evaluation and cost analysis, and the economics of education. In particular, he studies career and technical education, educational accountability policies, and the application of regression discontinuity research designs. In all of this work, he emphasizes how education can address human capital development as well as issues of equity related to race, class, gender, and disability.

References

Abadie

Imbens

G. W.

(2006). Large sample properties of matching estimators for average treatment effects. Econometrica, 74(1), 235–267.

Advance CTE. (2018). Career clusters. https://www.careertech.org/career-clusters

Altonji

Elder

Taber

(2005). Selection on observed and unobserved variables: Assessing the effectiveness of Catholic schools. Journal of Political Economy, 113(1), 151–184.

Anderson

(1982). The historical development of black vocational education. In Kantor

Tyack

D. B.

(Eds.), Work, youth and schooling: Historical perspectives on vocational education (pp. 180–222). Stanford University Press.

Ansel

Dougherty

Ecton

Holden

Theobald

(2022). Student interest and equitable access to career and technical education high schools of choice. Journal of School Choice. Advance online publication. https://doi.org/10.1080/15582159.2022.2039427

Autor

(2019). Work of the past, work of the future. National Bureau of Economic Research.

Belfield

Bailey

(2017). The labor market returns to sub-baccalaureate college: A review [A CAPSEE working paper]. Center for Analysis of Postsecondary Education and Employment.

Bishop

Mane

(2004). The impacts of career-technical education on high school labor market success. Economics of Education Review, 23(4), 381–402.

Böckerman

Haapanen

Jepsen

(2019). Back to school: Labor-market returns to higher vocational schooling. Labour Economics, 61, Article 101758.

10.

Bonilla

(2020). The dropout effects of career pathways: Evidence from California. Economics of Education Review, 75, Article 101972.

11.

Bowles

Gintis

(1976). Schooling in capitalist America. Basic Books.

12.

Bozick

Dalton

(2013). Balancing career and technical education with academic coursework: The consequences for mathematics achievement in high school. Educational Evaluation and Policy Analysis, 35(2), 123–138.

13.

Brunello

Rocco

(2017). The effects of vocational education on adult skills, employment and wages: What can we learn from PIAAC? SERIEs, 8(4), 315–343.

14.

Brunner

Dougherty

S. M.

Ross

(2021). The effects of career and technical education: Evidence from the Connecticut technical high school system (No. w28790). National Bureau of Economic Research.

15.

Carnevale

Jayasundera

Gulish

(2015). Good jobs are back: College graduates are first in line. Center on Education and the Workforce, Georgetown University. http://hdl.handle.net/10822/1050310

16.

Cellini

(2006). Smoothing the transition to college? The effect of tech-ed programs on educational attainment. Economics of Education Review, 25, 304–411.

17.

Collins

Garin

Jackson

Koustas

Payne

(2019). Is gig work replacing traditional employment? Evidence from two decades of tax returns [Unpublished paper, IRS SOI joint statistical research program].

18.

Dougherty

S. M.

(2016). Career and technical education in high school: Does it improve student outcomes? Thomas B. Fordham Institute.

19.

Dougherty

S. M.

(2018). The effect of career and technical education on human capital accumulation: Causal evidence from Massachusetts. Education Finance and Policy, 13(2), 119–148.

20.

Dougherty

S. M.

Ecton

W. G.

(2021). The economic effect of vocational education on student outcomes. In Oxford research encyclopedia of economics and finance. Oxford University Press.

21.

Dougherty

S. M.

Grindal

Hehir

(2018). The impact of career and technical education on students with disabilities. Journal of Disability Policy Studies, 29(2), 108–118.

22.

Dougherty

S. M.

Harbaugh Macdonald

(2020). Can growth in the availability of STEM technical education improve equality in participation? Evidence from Massachusetts. Journal of Vocational Education & Training, 72(1), 47–70.

23.

Foote

Stange

(2019). Attrition from administrative data: Problems and solutions with an application to higher education [Working paper, NBER education program meeting]. https://conference.nber.org/conf_papers/f131000.pdf

24.

Gottfried

Plasman

(2018). Linking the timing of career and technical education coursetaking with high school dropout and college-going behavior. American Educational Research Journal, 55(2), 325–361.

25.

Hanushek

Schwerdt

Woessmann

Zhang

(2017). General education, vocational education, and labor-market outcomes over the life-cycle. Journal of Human Resources, 52(1), 49–88.

26.

Hemelt

Lenard

Paeplow

(2019). Building bridges to life after high school: Contemporary career academies and student outcomes. Economics of Education Review, 68, 161–178.

27.

Holzer

Baum

(2017). Making college work: Pathways to success for disadvantaged students. Brookings Institution Press.

28.

Jepsen

Troske

Coomes

(2014). The labor-market returns to community college degrees, diplomas, and certificates. Journal of Labor Economics, 32(1), 95–121.

29.

Kelly

Price

(2009). Vocational education: A clear slate for disengaged students? Social Science Research, 38(4), 810–825.

30.

Kemple

Willner

(2008). Career academies: Long-term impacts on labor-market outcomes, educational attainment, and transitions to adulthood. Manpower Demonstration Research Corporation.

31.

Kreisman

Stange

(2020). Vocational and career tech education in American high schools: The value of depth over breadth. Education Finance and Policy, 15(1), 11–44.

32.

Lazear

E. P.

(2009). Firm-specific human capital: A skill-weights approach. Journal of Political Economy, 117(5), 914–940.

33.

Markus

(2011, February 3). Duncan joins calls to reinvent career technical education. Edutopia. https://www.edutopia.org/blog/arne-duncan-career-technical-education

34.

Meer

(2007). Evidence on the returns to secondary vocational education. Economics of Education Review, 26(5), 559–573.

35.

Oakes

(1983). Limiting opportunity: Student race and curricular differences in secondary vocational education. American Journal of Education, 91, 328–355.

36.

Oster

(2019). Unobservable selection and coefficient stability: Theory and evidence. Journal of Business & Economic Statistics, 37(2), 187–204.

37.

Plasman

Gottfried

Klasik

(2020). Trending up: A cross-cohort exploration of STEM career and technical education participation by low-income students. Journal of Education for Students Placed at Risk, 25(1), 55–78.

38.

Plasman

Gottfried

Sublett

(2017). Are the academic CTE cluster pipeline? Linking high school CTE coursetaking and postsecondary credentials. Career and Technical Education Research, 42(3), 219–242.

39.

Rosenbaum

P. R.

Rubin

D. B.

(1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1), 41–55.

40.

Stevens

A. H.

Kurlaender

Grosz

(2019). Career technical education and labor market outcomes evidence from California community colleges. Journal of Human Resources, 54(4), 986–1036.

41.

Tyack

(1974). The one best system: A history of American urban education. Harvard University Press.

42.

U.S. Department of Education. (2019). Bridging the skills gap: Career and technical education in high school. https://www2.ed.gov/datastory/cte/index.html

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

8.28 MB